B1643
Title: A performance evaluation of randomization methods in small-size clinical trials with binary or survival outcomes
Authors: Kanae Takahashi - Hyogo Medical University (Japan) [presenting]
Kouji Yamamoto - Yokohama City University (Japan)
Abstract: Clinical trials are the most definitive method for determining the treatment effect. The important component of clinical trials is randomization which is a technique used for assigning patients to either the experimental treatment(s) group or the control group. The selection of a randomization method is important at the time of protocol planning. There are several methods of randomization. Complete randomization, permuted block design, stratified blocked randomization, and minimization are relatively commonly used. The performance of randomization methods may be related to trial sample size, the setting of prognostic factors, and statistical analysis methods. However, for small-size randomized controlled trials with binary or survival outcomes, the impact of using a prognostic factor that has an interaction or a baseline characteristic that has no effect on the outcome for randomization and statistical analysis was not considered well in previous research. Thus, the objective is to compare the performance of the existing randomization methods when using a prognostic factor that has an interaction or a baseline characteristic that has no effect on the outcome by simulations and to suggest the recommended randomization method for small-size randomized controlled trials with binary or survival outcomes.